Industry experience in using an abstract model to select software development tools
Gertjan Zwartjes, van J Geffen, DG Kourie, Andrew Boake, Bruce W. Watson · TU/e Research Portal · 2004
Experience shows that it is difficult to select, from an ever-widening set of software tools, those appropriate for a given project.Complicating the choice beyond mere vendor selection is a proliferation of Open Source offerings.An abstract model is proposed to assist in this selection process.In this model, generic elements of a rational decision making process are identified.The model is realized in a matrix in which rows represent tools and columns represent tool properties.This tool matrix is developed iteratively, starting off with an initial attempt to articulate those tool properties perceived to be important in the project.The criticality of each property is also estimated.This information constitutes the column headings.Tools described in available information sources constitute the rows, with their compliance to the required properties indicated in the appropriate cells of the matrix.Rows are grouped together in functional categories -e.g.version systems, modeling tools, etc. Columns are grouped in terms of properties typically associated with a given functional category.The result is a visual representation of the extent to which identified tools meet properties.This serves as a basis for selecting one or more tools to meet the desired requirements.Iteration of the model consists of evaluating the coverage of these requirements, refinement of properties and trade off between alternatives.Experience in using the model in a number of industrial contexts is discussed.The case studies show that the process forces one to consider as many alternatives and prerequisites as are necessary, and to systematically make your reasoning explicit.This experience leads to the conclusion that the discipline demanded to construct the model has rational decisions as its payoff, based on well-assimilated information summarized in the tool matrix.